Building a dictionary of lexical variants for human phenotype descriptors

Abstract

Detecting phenotype descriptors in text and linking them to ontology concepts is a challenging task. Current state-of-the art concept recognizers struggle with several issues due the variety of human expressiveness. Here we present initial results of creating a dictionary of lexical variants for the Human Phenotype Ontology. This work is a smaller but important part of a larger project with a goal to improve recall in phenotype concept recognizers.